Skip to content

Error when compiling with CUDA 12 #17

@jcgarciaca

Description

@jcgarciaca

Hi,

I try to build the python wrapper with running ./compile_with_cuda.sh, but I get the following error:

error: Command "/usr/local/cuda-12.6/bin/nvcc -I../ -I/usr/lib/python3/dist-packages/numpy/core/include -I/usr/local/cuda-12.6/include -I/usr/include/python3.10 -c ../vendor/lodepng/lodepng.cpp -o build/temp.linux-aarch64-3.10/range_libc/vendor/lodepng/lodepng.o -w -std=c++11 -march=native -ffast-math -fno-math-errno -O3 -DUSE_CUDA=1 -DCHUNK_SIZE=262144 -DNUM_THREADS=256" failed with exit status 1

If I run that command I get

/usr/local/cuda-12.6/bin/nvcc -I../ -I/usr/lib/python3/dist-packages/numpy/core/include -I/usr/local/cuda-12.6/include -I/usr/include/python3.10 -c ../vendor/lodepng/lodepng.cpp -o build/temp.linux-aarch64-3.10/range_libc/vendor/lodepng/lodepng.o -w -std=c++11 -march=native -ffast-math -fno-math-errno -O3 -DUSE_CUDA=1 -DCHUNK_SIZE=262144 -DNUM_THREADS=256 nvcc fatal : Unknown option '-march=native'

In generated RangeLibc.cpp, I can see compile arguments seem correct (-march=native is not part of nvcc args)

        "extra_compile_args": {
            "gcc": [
                "-w",
                "-std=c++11",
                "-march=native",
                "-ffast-math",
                "-fno-math-errno",
                "-O3",
                "-DUSE_CUDA=1",
                "-DCHUNK_SIZE=262144",
                "-DNUM_THREADS=256"
            ],
            "nvcc": [
                "-arch=sm_87",
                "--ptxas-options=-v",
                "-c",
                "--compiler-options",
                "'-fPIC'",
                "-w",
                "-std=c++11",
                "-DUSE_CUDA=1",
                "-DCHUNK_SIZE=262144",
                "-DNUM_THREADS=256"
            ]
        },

And my nvcc version

nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Wed_Aug_14_10:14:07_PDT_2024
Cuda compilation tools, release 12.6, V12.6.68
Build cuda_12.6.r12.6/compiler.34714021_0

Any idea how to fix this issue?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions